[Drag0nFly]

My bad. I did a clean checkout of your branch, but it appears it had the wrong WinSystemX11.cpp with the (superfluous) CheckDisplayEvents() function in it. (The commit only affected one file, according to GitHub.)

---

That should be WinSystemX11GL.cpp ...

[FernetMenta]

Not sure what you are saying. My master compiles/links fine here.

[Drag0nFly]

See the previous comment. (I did a git pull of your branch, but it had the wrong WinSystemX11GL.cpp, when comparing with the one on GitHub.)

[FernetMenta]

My repo gets force pushed, hence git pull does not work. you need git fetch and git reset --hard

[Drag0nFly]

It compiled fine now after I replaced WinSystemX11GL.cpp

But more importantly:

Tested it and - gasp! - it works perfectly; holy sheepsh$t. Started right up on the correct port (DP2). Traversed the outputs, and it correctly reverted back to VGA1 and to DP2 when changing the output config.

I guess Chris Wilson @Intel (who is digging into the x11/i965 driver) might have an easier job ahead of him... Thanks again for fixing!

---

btw.-just saw your comment on git fetch & git reset --hard; I'll have to remember that next time.

[fritsch]

Also make sure that ccache is cleaned, which also has produced issues in the past.

Fernet's tree is "force pushed" tree - you won't get lucky with doing normal git pulls - always fetch the remote, rebase (if you have own work) or reset --hard if you don't

and copying files in the age of git should be avoided ...

[Drag0nFly]

@FernetMenta–you might be interested in Chris Wilson's comments in the Intel bug report: https://bugs.freedesktop.org/show_bug.cgi?id=82619#c43

He's been doing some testing with recently compiled versions that include the XMoveWindow patch and those preceding it; which suffered from the bug.

[FernetMenta]

See my comments on the bug report. Chris Wilson's proposed changes do not work.

[FernetMenta]

@DragonFly could you please test my current master branch.

[Drag0nFly]

Tested the latest changes now; XBMC starts up on the correct port (DP2), but as soon as you try to change outputs it briefly (~1 sec) switches to VGA1, then immediately reverts back again to DP2. This occurs regardless of which port is set to be the primary with xrandr.

Sorry for the late reply, btw.! I was away yesterday (otherwise I would have tested Chris' patches as well.) He does not seem entirely convinced that this is something that has been introduced on the driver/X side, but as you state it is too much of a coincidence that all WMs are currently failing.

I also experienced the same issues you did with the resolution mismatch (the XBMC window inheriting the other outputs' screen size), from comment #47.